Probability Distributions

Total Page:16

File Type:pdf, Size:1020Kb

Probability Distributions Probability distributions [Nematrian website page: ProbabilityDistributionsIntro, © Nematrian 2020] The Nematrian website contains information and analytics on a wide range of probability distributions, including: Discrete (univariate) distributions - Bernoulli, see also binomial distribution - Binomial - Geometric, see also negative binomial distribution - Hypergeometric - Logarithmic - Negative binomial - Poisson - Uniform (discrete) Continuous (univariate) distributions - Beta - Beta prime - Burr - Cauchy - Chi-squared - Dagum - Degenerate - Error function - Exponential - F - Fatigue, also known as the Birnbaum-Saunders distribution - Fréchet, see also generalised extreme value (GEV) distribution - Gamma - Generalised extreme value (GEV) - Generalised gamma - Generalised inverse Gaussian - Generalised Pareto (GDP) - Gumbel, see also generalised extreme value (GEV) distribution - Hyperbolic secant - Inverse gamma - Inverse Gaussian - Johnson SU - Kumaraswamy - Laplace - Lévy - Logistic - Log-logistic - Lognormal - Nakagami - Non-central chi-squared - Non-central t - Normal - Pareto - Power function - Rayleigh - Reciprocal - Rice - Student’s t - Triangular - Uniform - Weibull, see also generalised extreme value (GEV) distribution Continuous multivariate distributions - Inverse Wishart Copulas (a copula is a special type of continuous multivariate distribution) - Clayton - Comonotonicity - Countermonotonicity (only valid for 푛 = 2, where 푛 is the dimension of the input) - Frank - Generalised Clayton - Gumbel - Gaussian - Independence - t The Nematrian website functions for fitting univariate distributions, creating random variates and calculating moments etc. now cover most of the above probability distributions, see ProbabilityDistributionsFunctions. In many cases these functions can cater for the traditional (textbook) forms of these distributions and variants that include additional shift and scale parameters. location (i.e. shift) and scale variants [ProbabilityDistributionsIntro2] The location and scale of any probability distribution can be adjusted by using the (linear) transform 푌 = 푔 + ℎ푋 where 푔 and ℎ are constants (푔 adjusts the location, i.e. shifts the distribution, whilst ℎ adjusts its scale). This leaves the skew and (excess kurtosis) unaltered but alters the mean and variance as 퐸(푌) = 푔 + ℎ퐸(푋) and 푣푎푟(푌) = ℎ2푣푎푟(푋). In some cases the typical distributional specification already includes such components. For example, the normal distribution 푁(휇, 2) is the location and scale adjusted version of the unit normal distribution 푁(0,1). In other cases the standard distributional specification does not include such adjustments. For example, the (standard) Student’s t distribution depends one just one parameter, its degrees of freedom. The probability distribution orientated Nematrian web functions recognise location and/or scale adjusted variants of wide range of standard probability distributions. Discrete (univariate) distributions The Bernoulli distribution [BernoulliDistribution] A Bernoulli trial is an experiment that has one of two possible outcomes, ‘success’ with probability 푝 and ‘failure’ with probability 1 − 푝. The Bernoulli distribution is a probability distribution that takes the value of 1 if such a trial is a ‘success’ and 0 if it is a ‘failure’. The Bernoulli distribution is a special case of the binomial distribution, 퐵(푛, 푝), with 푛 = 1. For characteristics of the Bernoulli distribution (e.g. mean, standard deviation etc.), please refer to the corresponding characteristics for the binomial distribution. For other probability distributions see here. The binomial distribution [BinomialDistribution] The binomial distribution 퐵(푛, 푝) is the discrete probability distribution applicable to the number of successes in a sequence of 푛 independent yes/no experiments each of which has a success probability of 푝. Each individual success/failure experiment is called a Bernoulli trial, so if 푛 = 1 then the binomial distribution is a Bernoulli distribution. It has the following characteristics: Distribution name Binomial distribution Common notation 푋~퐵(푛, 푝) Parameters 푛 = number of (independent) trials, positive integer 푝 = probability of success in each trial, 0 ≤ 푝 ≤ 1 Support 푥 ∈ {0,1, … , 푛} = 푛푢푚푏푒푟 표푓 푠푢푐푐푒푠푠푒푠 Probability mass function 푛 푛! 푓(푥) = ( ) 푝푥(1 − 푝)푛−푥 = 푝푥(1 − 푝)푛−푥 푥 (푛 − 푥)! 푥! Cumulative distribution 푥 푛 푗 푛−푗 function 퐹(푥) = ∑ (푗 ) 푝 (1 − 푝) = 퐼1−푝(푛 − 푥, 푥 + 1) 푗=0 Mean 푛푝 Variance 푛푝(1 − 푝) Skewness 1 − 2푝 √푛푝(1 − 푝) (Excess) kurtosis 1 − 6푝(1 − 푝) 푛푝(1 − 푝) 푛 Characteristic function (1 − 푝 + 푝푒푖푡) Other comments The Bernoulli distribution is 퐵(1, 푝) and corresponds to the likelihood of success of a single experiment. Its probability mass function and cumulative distribution function are: 1 − 푝, 푥 = 0 푓(푥) = 퐹(푥) = { 푝, 푥 = 1 The Bernoulli distribution with 푝 = 1⁄2, i.e. 퐵(1, 1⁄2), has the minimum possible excess kurtosis, i.e. −2. The mode of 퐵(푛, 푝) is 푖푛푡((푛 + 1)푝) if (푛 + 1)푝 is 0 or not an integer and is 푛 if (푛 + 1)푝 = 푛 + 1. If (푛 + 1)푝 ∈ {1,2, … , 푛} then the distribution is bi-modal, with modes (푛 + 1)푝 and (푛 + 1)푝 − 1. The binomial distribution is often used to model the number of successes in a sample size of 푛 from a population size of 푁. Since such samples are not independent, the resulting distribution is actually a hypergeometric distribution and not a binomial distribution. However if 푁 ≫ 푛 then the binomial distribution becomes a good approximation to the relevant hypergeometric distribution and is thus often used. 푛 In the above ( ) is the binomial coefficient. 푥 Nematrian web functions Functions relating to the above distribution may be accessed via the Nematrian web function library by using a DistributionName of “binomial”. For details of other supported probability distributions see here. The geometric distribution [GeometricDistribution] The geometric distribution describes the probability of 푥 successes in a sequence of independent experiments each with likelihood of success of 푝 that arise before there is 1 failure. It is a special case of the negative binomial distribution. Note: different texts adopt slightly different definitions, e.g. it may be the total number of trials (i.e. 1 more than the above) in which case it may be called the shifted geometric distribution and/or 푝 may denote the probability of failure rather than the probability of success. The hypergeometric distribution [HypergeometricDistribution] The hypergeometric distribution describes the probability of 푥 successes in 푛 draws from a finite population size 푁 containing 푚 successes without replacement. This contrasts with the binomial distribution which describes the probability of 푥 successes in 푛 draws with replacement Distribution name Hypergeometric distribution Common notation 푋~퐻푦푝푒푟푔푒표푚푒푡푟푖푐(푚, 푁, 푛) Parameters 푁 = population size, integral (푁 > 0) 푛 = sample size, integral (0 < 푛 ≤ 푁) 푚 = number of tagged items, integral (0 < 푚 ≤ 푁) Domain max(0, 푛 + 푚 − 푁) ≤ 푥 ≤ max(푛, 푚) , 푥 an integer 푚 푁 − 푚 Probability mass function ( ) ( ) 푥 푓(푥) = 푛 − 푥 푁 ( ) 푛 Cumulative distribution 푚 푁 − 푚 푛 푁 − 푛 푥 ( ) ( ) ( ) ( ) 푗 푛 − 푗 푘 + 1 function 퐹(푥) = (푥) = ∑ = 1 − 푚 − 푘 − 1 푌 푁 푁 ( ) ( ) 푗=0 푗 푚 where 푌 = 3퐹2(1, 푥 + 1 − 푚, 푥 + 1 − 푛; 푥 + 2, 푁 + 푥 + 2 − 푚 − 푛; 1) 푝퐹푞 is the generalised hypergeometric function, i.e. ∞ (푎 ) … (푎 ) 푛 1 푛 푝 푛 푧 푝퐹푞(푎1, … , 푎푝; 푏1, … , 푏푞; 푧) = ∑ (푏 ) … (푏 ) 푛! 푛=0 1 푛 푞 푛 and (푎)푛 involves the rising factorial or Pochhammer notation, i.e. (푎)푛 = 푎(푎 + 1)(푎 + 2) … (푎 + 푛 − 1) and (푎)0 = 1 푛푚 Mean 푁 Variance 푛푚(푁 − 푚)(푁 − 푛) 푁2(푁 − 1) Skewness (푁 − 2푚)(푁 − 1)1⁄2(푁 − 2푛) 1⁄2 (푛푚(푁 − 푚)(푁 − 푛)) (푁 − 2) (Excess) kurtosis 퐴 + 퐵 퐶 where 퐴 = (푁 − 1)푁2(푁(푁 + 1) − 6푚(푁 − 푚) − 6푛(푁 − 푚)) 퐵 = 6푛푚(푁 − 푚)(푁 − 푛)(5푁 − 6) 퐶 = 푛푚(푁 − 푚)(푁 − 푛)(푁 − 2)(푁 − 3) 푁 − 푚 Characteristic function ( ) 푛 퐹 (−푛, −푚; 푛 − 푚 − 푛 + 1; 푒푖푡) 푁 2 1 ( ) 푛 Nematrian web functions Functions relating to the above distribution may be accessed via the Nematrian web function library by using a DistributionName of “hypergeometric”. For details of other supported probability distributions see here. The logarithmic distribution [LogarithmicDistribution] The logarithmic distribution arises from following power series expansion: 푝2 푝3 − log(1 − 푝) = 푝 + + + ⋯ 2 3 푝푥 This means that the function 푓(푥) = − , 푥 = 1,2,3, … can naturally be interpreted as a 푥 log(1−푝) ∞ probability mass function since ∑푘=1 푓(푘) = 1. Distribution name Logarithmic distribution Common notation 푋~퐿표푔(푝) Parameters 푝 = shape parameter (0 < 푝 < 1) Domain 1 ≤ 푥 < +∞, 푥 an integer Probability mass function 푝푥 푓(푥) = − 푥 log(1 − 푝) Cumulative distribution 푥 푗 1 푝 퐵푝(푥 + 1,0) 퐹(푥) = − ∑ = 1 + function log(1 − 푝) 푗 log(1 − 푝) 푗=1 푝 Mean − (1 − 푝) log(1 − 푝) Variance 푝(푝 + log(1 − 푝)) − = 푉 (1 − 푝)2(log(1 − 푝))2 Skewness 푝 3푝 2푝2 − (1 + 푝 + + ) (1 − 푝)3푉3⁄2 log(1 − 푝) log(1 − 푝) log2(1 − 푝) 푝 (Excess) kurtosis − 퐴 − 3 (1 − 푝)4푉2 log(1 − 푝) where 4푝(1 + 푝) 6푝2 3푝3 퐴 = (1 + 4푝 + 푝2 + + + ) log(1 − 푝) log2(1 − 푝) log3(1 − 푝) Characteristic function log(1 − 푝푒푖푡) log(1 − 푝) Other comments The logarithmic distribution has a mode of 1. If 푁 is a random variable with Poission distribution and 푋푖, 푖 = 1, … is an infinite sequence of iid random variables each distributed 퐿표푔(푝) then 푌 = 푁 ∑푖=1 푋푖 has a negative binomial distribution showing that the negative binomial distribution is an example of a compound Poisson distribution Nematrian web functions Functions relating to the above distribution may be accessed via the Nematrian web function library by using a DistributionName of “logarithmic”. For details of other supported probability distributions see here. The negative binomial distribution [NegativeBinomialDistribution] The negative binomial distribution describes the probability of 푥 successes in a sequence of independent experiments each with likelihood of success of 푝 that arise before there are 푟 failures.
Recommended publications
  • Package 'Prevalence'
    Zurich Open Repository and Archive University of Zurich Main Library Strickhofstrasse 39 CH-8057 Zurich www.zora.uzh.ch Year: 2013 Package ‘prevalence’ Devleesschauwer, Brecht ; Torgerson, Paul R ; Charlier, Johannes ; Levecke, Bruno ; Praet, Nicolas ; Dorny, Pierre ; Berkvens, Dirk ; Speybroeck, Niko Abstract: Tools for prevalence assessment studies. IMPORTANT: the truePrev functions in the preva- lence package call on JAGS (Just Another Gibbs Sampler), which therefore has to be available on the user’s system. JAGS can be downloaded from http://mcmc-jags.sourceforge.net/ Posted at the Zurich Open Repository and Archive, University of Zurich ZORA URL: https://doi.org/10.5167/uzh-89061 Scientific Publication in Electronic Form Published Version The following work is licensed under a Software: GNU General Public License, version 2.0 (GPL-2.0). Originally published at: Devleesschauwer, Brecht; Torgerson, Paul R; Charlier, Johannes; Levecke, Bruno; Praet, Nicolas; Dorny, Pierre; Berkvens, Dirk; Speybroeck, Niko (2013). Package ‘prevalence’. On Line: The Comprehensive R Archive Network. Package ‘prevalence’ September 22, 2013 Type Package Title The prevalence package Version 0.2.0 Date 2013-09-22 Author Brecht Devleesschauwer [aut, cre], Paul Torgerson [aut],Johannes Charlier [aut], Bruno Lev- ecke [aut], Nicolas Praet [aut],Pierre Dorny [aut], Dirk Berkvens [aut], Niko Speybroeck [aut] Maintainer Brecht Devleesschauwer <[email protected]> BugReports https://github.com/brechtdv/prevalence/issues Description Tools for prevalence
    [Show full text]
  • A Matrix-Valued Bernoulli Distribution Gianfranco Lovison1 Dipartimento Di Scienze Statistiche E Matematiche “S
    Journal of Multivariate Analysis 97 (2006) 1573–1585 www.elsevier.com/locate/jmva A matrix-valued Bernoulli distribution Gianfranco Lovison1 Dipartimento di Scienze Statistiche e Matematiche “S. Vianelli”, Università di Palermo Viale delle Scienze 90128 Palermo, Italy Received 17 September 2004 Available online 10 August 2005 Abstract Matrix-valued distributions are used in continuous multivariate analysis to model sample data matrices of continuous measurements; their use seems to be neglected for binary, or more generally categorical, data. In this paper we propose a matrix-valued Bernoulli distribution, based on the log- linear representation introduced by Cox [The analysis of multivariate binary data, Appl. Statist. 21 (1972) 113–120] for the Multivariate Bernoulli distribution with correlated components. © 2005 Elsevier Inc. All rights reserved. AMS 1991 subject classification: 62E05; 62Hxx Keywords: Correlated multivariate binary responses; Multivariate Bernoulli distribution; Matrix-valued distributions 1. Introduction Matrix-valued distributions are used in continuous multivariate analysis (see, for exam- ple, [10]) to model sample data matrices of continuous measurements, allowing for both variable-dependence and unit-dependence. Their potentials seem to have been neglected for binary, and more generally categorical, data. This is somewhat surprising, since the natural, elementary representation of datasets with categorical variables is precisely in the form of sample binary data matrices, through the 0-1 coding of categories. E-mail address: [email protected] URL: http://dssm.unipa.it/lovison. 1 Work supported by MIUR 60% 2000, 2001 and MIUR P.R.I.N. 2002 grants. 0047-259X/$ - see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.jmva.2005.06.008 1574 G.
    [Show full text]
  • Simulation of Moment, Cumulant, Kurtosis and the Characteristics Function of Dagum Distribution
    264 IJCSNS International Journal of Computer Science and Network Security, VOL.17 No.8, August 2017 Simulation of Moment, Cumulant, Kurtosis and the Characteristics Function of Dagum Distribution Dian Kurniasari1*,Yucky Anggun Anggrainy1, Warsono1 , Warsito2 and Mustofa Usman1 1Department of Mathematics, Faculty of Mathematics and Sciences, Universitas Lampung, Indonesia 2Department of Physics, Faculty of Mathematics and Sciences, Universitas Lampung, Indonesia Abstract distribution is a special case of Generalized Beta II The Dagum distribution is a special case of Generalized Beta II distribution. Dagum (1977, 1980) distribution has been distribution with the parameter q=1, has 3 parameters, namely a used in studies of income and wage distribution as well as and p as shape parameters and b as a scale parameter. In this wealth distribution. In this context its features have been study some properties of the distribution: moment, cumulant, extensively analyzed by many authors, for excellent Kurtosis and the characteristics function of the Dagum distribution will be discussed. The moment and cumulant will be survey on the genesis and on empirical applications see [4]. discussed through Moment Generating Function. The His proposals enable the development of statistical characteristics function will be discussed through the expectation of the definition of characteristics function. Simulation studies distribution used to empirical income and wealth data that will be presented. Some behaviors of the distribution due to the could accommodate both heavy tails in empirical income variation values of the parameters are discussed. and wealth distributions, and also permit interior mode [5]. Key words: A random variable X is said to have a Dagum distribution Dagum distribution, moment, cumulant, kurtosis, characteristics function.
    [Show full text]
  • 5.1 Convergence in Distribution
    556: MATHEMATICAL STATISTICS I CHAPTER 5: STOCHASTIC CONVERGENCE The following definitions are stated in terms of scalar random variables, but extend naturally to vector random variables defined on the same probability space with measure P . Forp example, some results 2 are stated in terms of the Euclidean distance in one dimension jXn − Xj = (Xn − X) , or for se- > quences of k-dimensional random variables Xn = (Xn1;:::;Xnk) , 0 1 1=2 Xk @ 2A kXn − Xk = (Xnj − Xj) : j=1 5.1 Convergence in Distribution Consider a sequence of random variables X1;X2;::: and a corresponding sequence of cdfs, FX1 ;FX2 ;::: ≤ so that for n = 1; 2; :: FXn (x) =P[Xn x] : Suppose that there exists a cdf, FX , such that for all x at which FX is continuous, lim FX (x) = FX (x): n−!1 n Then X1;:::;Xn converges in distribution to random variable X with cdf FX , denoted d Xn −! X and FX is the limiting distribution. Convergence of a sequence of mgfs or cfs also indicates conver- gence in distribution, that is, if for all t at which MX (t) is defined, if as n −! 1, we have −! () −!d MXi (t) MX (t) Xn X: Definition : DEGENERATE DISTRIBUTIONS The sequence of random variables X1;:::;Xn converges in distribution to constant c if the limiting d distribution of X1;:::;Xn is degenerate at c, that is, Xn −! X and P [X = c] = 1, so that { 0 x < c F (x) = X 1 x ≥ c Interpretation: A special case of convergence in distribution occurs when the limiting distribution is discrete, with the probability mass function only being non-zero at a single value, that is, if the limiting random variable is X, then P [X = c] = 1 and zero otherwise.
    [Show full text]
  • Exponentiated Kumaraswamy-Dagum Distribution with Applications to Income and Lifetime Data Shujiao Huang and Broderick O Oluyede*
    Huang and Oluyede Journal of Statistical Distributions and Applications 2014, 1:8 http://www.jsdajournal.com/content/1/1/8 RESEARCH Open Access Exponentiated Kumaraswamy-Dagum distribution with applications to income and lifetime data Shujiao Huang and Broderick O Oluyede* *Correspondence: [email protected] Abstract Department of Mathematical A new family of distributions called exponentiated Kumaraswamy-Dagum (EKD) Sciences, Georgia Southern University, Statesboro, GA 30458, distribution is proposed and studied. This family includes several well known USA sub-models, such as Dagum (D), Burr III (BIII), Fisk or Log-logistic (F or LLog), and new sub-models, namely, Kumaraswamy-Dagum (KD), Kumaraswamy-Burr III (KBIII), Kumaraswamy-Fisk or Kumaraswamy-Log-logistic (KF or KLLog), exponentiated Kumaraswamy-Burr III (EKBIII), and exponentiated Kumaraswamy-Fisk or exponentiated Kumaraswamy-Log-logistic (EKF or EKLLog) distributions. Statistical properties including series representation of the probability density function, hazard and reverse hazard functions, moments, mean and median deviations, reliability, Bonferroni and Lorenz curves, as well as entropy measures for this class of distributions and the sub-models are presented. Maximum likelihood estimates of the model parameters are obtained. Simulation studies are conducted. Examples and applications as well as comparisons of the EKD and its sub-distributions with other distributions are given. Mathematics Subject Classification (2000): 62E10; 62F30 Keywords: Dagum distribution; Exponentiated Kumaraswamy-Dagum distribution; Maximum likelihood estimation 1 Introduction Camilo Dagum proposed the distribution which is referred to as Dagum distribution in 1977. This proposal enable the development of statistical distributions used to fit empir- ical income and wealth data, that could accommodate heavy tails in income and wealth distributions.
    [Show full text]
  • Discrete Probability Distributions Uniform Distribution Bernoulli
    Discrete Probability Distributions Uniform Distribution Experiment obeys: all outcomes equally probable Random variable: outcome Probability distribution: if k is the number of possible outcomes, 1 if x is a possible outcome p(x)= k ( 0 otherwise Example: tossing a fair die (k = 6) Bernoulli Distribution Experiment obeys: (1) a single trial with two possible outcomes (success and failure) (2) P trial is successful = p Random variable: number of successful trials (zero or one) Probability distribution: p(x)= px(1 − p)n−x Mean and variance: µ = p, σ2 = p(1 − p) Example: tossing a fair coin once Binomial Distribution Experiment obeys: (1) n repeated trials (2) each trial has two possible outcomes (success and failure) (3) P ith trial is successful = p for all i (4) the trials are independent Random variable: number of successful trials n x n−x Probability distribution: b(x; n,p)= x p (1 − p) Mean and variance: µ = np, σ2 = np(1 − p) Example: tossing a fair coin n times Approximations: (1) b(x; n,p) ≈ p(x; λ = pn) if p ≪ 1, x ≪ n (Poisson approximation) (2) b(x; n,p) ≈ n(x; µ = pn,σ = np(1 − p) ) if np ≫ 1, n(1 − p) ≫ 1 (Normal approximation) p Geometric Distribution Experiment obeys: (1) indeterminate number of repeated trials (2) each trial has two possible outcomes (success and failure) (3) P ith trial is successful = p for all i (4) the trials are independent Random variable: trial number of first successful trial Probability distribution: p(x)= p(1 − p)x−1 1 2 1−p Mean and variance: µ = p , σ = p2 Example: repeated attempts to start
    [Show full text]
  • Benford's Law As a Logarithmic Transformation
    Benford’s Law as a Logarithmic Transformation J. M. Pimbley Maxwell Consulting, LLC Abstract We review the history and description of Benford’s Law - the observation that many naturally occurring numerical data collections exhibit a logarithmic distribution of digits. By creating numerous hypothetical examples, we identify several cases that satisfy Benford’s Law and several that do not. Building on prior work, we create and demonstrate a “Benford Test” to determine from the functional form of a probability density function whether the resulting distribution of digits will find close agreement with Benford’s Law. We also discover that one may generalize the first-digit distribution algorithm to include non-logarithmic transformations. This generalization shows that Benford’s Law rests on the tendency of the transformed and shifted data to exhibit a uniform distribution. Introduction What the world calls “Benford’s Law” is a marvelous mélange of mathematics, data, philosophy, empiricism, theory, mysticism, and fraud. Even with all these qualifiers, one easily describes this law and then asks the simple questions that have challenged investigators for more than a century. Take a large collection of positive numbers which may be integers or real numbers or both. Focus only on the first (non-zero) digit of each number. Count how many numbers of the large collection have each of the nine possibilities (1-9) as the first digit. For typical J. M. Pimbley, “Benford’s Law as a Logarithmic Transformation,” Maxwell Consulting Archives, 2014. number collections – which we’ll generally call “datasets” – the first digit is not equally distributed among the values 1-9.
    [Show full text]
  • 1988: Logarithmic Series Distribution and Its Use In
    LOGARITHMIC SERIES DISTRIBUTION AND ITS USE IN ANALYZING DISCRETE DATA Jeffrey R, Wilson, Arizona State University Tempe, Arizona 85287 1. Introduction is assumed to be 7. for any unit and any trial. In a previous paper, Wilson and Koehler Furthermore, for ~ach unit the m trials are (1988) used the generalized-Dirichlet identical and independent so upon summing across multinomial model to account for extra trials X. = (XI~. X 2 ...... X~.)', the vector of variation. The model allows for a second order counts ~r the j th ~nit has ~3multinomial of pairwise correlation among units, a type of distribution with probability vector ~ = (~1' assumption found reasonable in some biological 72 ..... 71) ' and sample size m. Howe~er, data. In that paper, the two-way crossed responses given by the J units at a particular generalized Dirichlet Multinomial model was used trial may be correlated, producing a set of J to analyze repeated measure on the categorical correlated multinomial random vectors, X I, X 2, preferences of insurance customers. The number • o.~ X • of respondents was assumed to be fixed and Ta~is (1962) developed a model for this known. situation, which he called the generalized- In this paper a generalization of the model multinomial distribution in which a single is made allowing the number of respondents m, to parameter P, is used to reflect the common be random. Thus both the number of units m, and dependency between any two of the dependent the underlying probability vector are allowed to multinomial random vectors. The distribution of vary. The model presented here uses the m logarithmic series distribution to account for the category total Xi3.
    [Show full text]
  • Parameter Specification of the Beta Distribution and Its Dirichlet
    %HWD'LVWULEXWLRQVDQG,WV$SSOLFDWLRQV 3DUDPHWHU6SHFLILFDWLRQRIWKH%HWD 'LVWULEXWLRQDQGLWV'LULFKOHW([WHQVLRQV 8WLOL]LQJ4XDQWLOHV -5HQpYDQ'RUSDQG7KRPDV$0D]]XFKL (Submitted January 2003, Revised March 2003) I. INTRODUCTION.................................................................................................... 1 II. SPECIFICATION OF PRIOR BETA PARAMETERS..............................................5 A. Basic Properties of the Beta Distribution...............................................................6 B. Solving for the Beta Prior Parameters...................................................................8 C. Design of a Numerical Procedure........................................................................12 III. SPECIFICATION OF PRIOR DIRICHLET PARAMETERS................................. 17 A. Basic Properties of the Dirichlet Distribution...................................................... 18 B. Solving for the Dirichlet prior parameters...........................................................20 IV. SPECIFICATION OF ORDERED DIRICHLET PARAMETERS...........................22 A. Properties of the Ordered Dirichlet Distribution................................................. 23 B. Solving for the Ordered Dirichlet Prior Parameters............................................ 25 C. Transforming the Ordered Dirichlet Distribution and Numerical Stability ......... 27 V. CONCLUSIONS........................................................................................................ 31 APPENDIX...................................................................................................................
    [Show full text]
  • The Kumaraswamy Pareto IV Distribution
    Austrian Journal of Statistics July 2021, Volume 50, 1{22. AJShttp://www.ajs.or.at/ doi:10.17713/ajs.v50i5.96 The Kumaraswamy Pareto IV Distribution M. H. Tahir Gauss M. Cordeiro M. Mansoor M. Zubair Ayman Alzaatreh Islamia University Federal University Govt. S.E. College Govt. S.E. College American University of Bahawapur of Pernambuco Bahawalpur Bahawalpur of Sharjah Abstract We introduce a new model named the Kumaraswamy Pareto IV distribution which extends the Pareto and Pareto IV distributions. The density function is very flexible and can be left-skewed, right-skewed and symmetrical shapes. It has increasing, decreasing, upside-down bathtub, bathtub, J and reversed-J shaped hazard rate shapes. Various structural properties are derived including explicit expressions for the quantile function, ordinary and incomplete moments, Bonferroni and Lorenz curves, mean deviations, mean residual life, mean waiting time, probability weighted moments and generating function. We provide the density function of the order statistics and their moments. The R´enyi and q entropies are also obtained. The model parameters are estimated by the method of maximum likelihood and the observed information matrix is determined. The usefulness of the new model is illustrated by means of three real-life data sets. In fact, our proposed model provides a better fit to these data than the gamma-Pareto IV, gamma-Pareto, beta-Pareto, exponentiated Pareto and Pareto IV models. Keywords: Arnold's Pareto, Kumaraswamy-G class, Pareto family, reliability. 1. Introduction The Pareto distribution and its generalizations are tractable statistical models for scien- tists, economists and engineers. These distributions cover a wide range of applications espe- cially in economics, finance, actuarial science, risk theory, reliability, telecommunications and medicine.
    [Show full text]
  • Severity Models - Special Families of Distributions
    Severity Models - Special Families of Distributions Sections 5.3-5.4 Stat 477 - Loss Models Sections 5.3-5.4 (Stat 477) Claim Severity Models Brian Hartman - BYU 1 / 1 Introduction Introduction Given that a claim occurs, the (individual) claim size X is typically referred to as claim severity. While typically this may be of continuous random variables, sometimes claim sizes can be considered discrete. When modeling claims severity, insurers are usually concerned with the tails of the distribution. There are certain types of insurance contracts with what are called long tails. Sections 5.3-5.4 (Stat 477) Claim Severity Models Brian Hartman - BYU 2 / 1 Introduction parametric class Parametric distributions A parametric distribution consists of a set of distribution functions with each member determined by specifying one or more values called \parameters". The parameter is fixed and finite, and it could be a one value or several in which case we call it a vector of parameters. Parameter vector could be denoted by θ. The Normal distribution is a clear example: 1 −(x−µ)2=2σ2 fX (x) = p e ; 2πσ where the parameter vector θ = (µ, σ). It is well known that µ is the mean and σ2 is the variance. Some important properties: Standard Normal when µ = 0 and σ = 1. X−µ If X ∼ N(µ, σ), then Z = σ ∼ N(0; 1). Sums of Normal random variables is again Normal. If X ∼ N(µ, σ), then cX ∼ N(cµ, cσ). Sections 5.3-5.4 (Stat 477) Claim Severity Models Brian Hartman - BYU 3 / 1 Introduction some claim size distributions Some parametric claim size distributions Normal - easy to work with, but careful with getting negative claims.
    [Show full text]
  • Package 'Distributional'
    Package ‘distributional’ February 2, 2021 Title Vectorised Probability Distributions Version 0.2.2 Description Vectorised distribution objects with tools for manipulating, visualising, and using probability distributions. Designed to allow model prediction outputs to return distributions rather than their parameters, allowing users to directly interact with predictive distributions in a data-oriented workflow. In addition to providing generic replacements for p/d/q/r functions, other useful statistics can be computed including means, variances, intervals, and highest density regions. License GPL-3 Imports vctrs (>= 0.3.0), rlang (>= 0.4.5), generics, ellipsis, stats, numDeriv, ggplot2, scales, farver, digest, utils, lifecycle Suggests testthat (>= 2.1.0), covr, mvtnorm, actuar, ggdist RdMacros lifecycle URL https://pkg.mitchelloharawild.com/distributional/, https: //github.com/mitchelloharawild/distributional BugReports https://github.com/mitchelloharawild/distributional/issues Encoding UTF-8 Language en-GB LazyData true Roxygen list(markdown = TRUE, roclets=c('rd', 'collate', 'namespace')) RoxygenNote 7.1.1 1 2 R topics documented: R topics documented: autoplot.distribution . .3 cdf..............................................4 density.distribution . .4 dist_bernoulli . .5 dist_beta . .6 dist_binomial . .7 dist_burr . .8 dist_cauchy . .9 dist_chisq . 10 dist_degenerate . 11 dist_exponential . 12 dist_f . 13 dist_gamma . 14 dist_geometric . 16 dist_gumbel . 17 dist_hypergeometric . 18 dist_inflated . 20 dist_inverse_exponential . 20 dist_inverse_gamma
    [Show full text]